Social learning theory (SLT) has been a useful model for understanding substance abuse. While approaches based on learning principles such as coping skills and relapse prevention have been promising, they have had varied results in recent empirical treatment studies involving alcohol and cocaine abuse. The proposed study is designed to develop and pilot a revised and refocused cognitive-behavioral psychotherapy approach for cocaine abusers, based on cocaine-related expectations. Expectancies appear to be a central construct that holds much promise for clinical research involving the assessment, etiology, and treatment of cocaine abuse. However, while other models do address expectation in some manner, expectancies have received little clinical treatment emphasis. Moreover, cocaine-related expectancies are readily measured using a validated, multifactorial, self-report instrument developed by the principle investigator of this study (Cocaine Expectancy Questionnaire, CEQ) and represent a domain that is amenable to change with talking therapy. For these reasons, we believe that expectancies have promising implications for treatment, that an intervention program centered around expectancies is warranted, and that such an intervention may substantially improve treatment efficacy. This study proposes to develop and assess the benefits of a manual driven, Expectancy-based Coping Skills Therapy (ECS) for the treatment of cocaine abuse. Instruments such as the CEQ, combined with such manualized interventions, make it possible to assess and use expectations about cocaine in a clearly specified treatment for cocaine abusers that can be reliably delivered, monitored, and evaluated.